291T Poster - Population Genetics
Thursday June 09, 9:15 PM - 10:00 PM

Authors:
David Peede ^{1, 2}; Diego Ortega-Del Vecchyo ³; Emilia Huerta-Sánchez ^{1, 2}

Affiliations:
1) Department of Ecology, Evolution, and Organismal Biology, Brown University, Providence, RI; 2) Center for Computational Molecular Biology, Brown University, Providence, RI; 3) Laboratorio Internacional de Investigación sobre el Genoma Humano, Universidad Nacional Autónoma de México, Juriquilla, Querétaro, México

Keywords:
Theory & Method Development

The evolution of species has traditionally been viewed as a bifurcating process. However, the recent influx of genomic studies has challenged this notion, demonstrating that hybridization is not only common in nature but is also a powerful force in shaping the evolution of genomes and patterns of genetic variation. Introgression’s role in the evolution of species has most often been inferred from whole-genome sequence data using summary statistics that measure the ratio of discordant phylogenetic relationships between gene trees and the assumed species tree by using site pattern frequencies as a proxy for gene tree frequencies. These methods mostly focus on measuring levels of shared derived alleles between the recipient and donor lineages to quantify the amount of introgression in the genome—also known as the admixture proportion. Introgression, however, also re-introduces ancestral alleles, and not including sites where the recipient and donor lineages share ancestral alleles leaves out sites that can be informative of introgression.
Here, we leverage both ancestral and derived allele sharing to 1) derive the analytical expectations of all possible topologies for a three-taxon species tree and subsequently derive the analytical expectations for existing estimators of the admixture proportion f_HOM, d_f, and D_P as a function of the mutation rate, effective population size, divergence times, the timing of introgression, and the admixture proportion; 2) define new summary statistics for quantifying the admixture proportion that leverage both ancestral and derived allele sharing and; 3) perform a simulation study where we vary the direction, timing, and amount of introgression, to assess if incorporating both patterns of derived and ancestral allele sharing improves estimates of the admixture proportion. We show that quantifying the admixture proportion is context-dependent, but in the absence of no a priori information about the direction, timing, or amount of introgression, our new estimator of the admixture proportion (P_D)—which considers both patterns of derived and ancestral allele sharing—performs the best 57% of the time across our 180 different demographic scenarios.